Nvidia has recently made a significant foray in simulation and GPU computing by deciding to open-source the source code relating to its PhysX SDK as well as the GPU compute shaders of the Flow SDK. This is a significant shift in the company’s past strategy. Even though the PhysX SDK had earlier been released under a BSD-3 license in 2018, the core of the system, i.e., the source code governing the GPU simulation kernel, had been kept tightly under wraps as Nvidia’s proprietary.

And now, in a sign of increased receptiveness to the tech community, Nvidia has opened access to over 500 CUDA kernels that are the basis of advanced capabilities like rigid body simulation with high-intensity interactions, fluid dynamics with their complicated currents, and realistic deformation of simulated objects. The goal of this project is ostensibly to create a living wave of innovation and foster an environment of collective learning among the vast universe of developers and researchers. By giving them unrestricted access and room for experimentation with such high-order simulation tools, Nvidia envisions a near-future revolution in the industry.

A solution to the problematic issue of real-time simulation

Alongside this notable opening, Nvidia has extended its policy of openness a further step by making access to the use of GPU compute shaders also available through its Flow SDK. This software library was designed specifically to offer a solution to the problematic issue of real-time simulation of gases and liquids. Its uniqueness is the use of sparse grids, a highly advanced technique for facilitating previously unimaginable degrees of accuracy and realism in simulations. With these advanced tools now available as open-source technology, programmers will now be able to solve extremely difficult computational problems in a broad array of applications ranging from simulating fluid behaviour to detailed simulations of physical interactions between a variety of virtual entities. This is a major leap forward for the entire developer community.

The personal motivation for the decision to donate these valuable resources to the community comes from the desire to foster fruitful collaboration between researchers active in the area and to provoke an acceleration of the innovation process within the very significant area of parallel computing and physical simulation. It is hoped that this strategic move will enable developers to overcome the current technical constraints that still hold back real-time simulations, thus opening new and promising avenues for significant breakthroughs in cutting-edge areas such as computer graphics, advanced robotics, and the precise simulation of complex natural phenomena.

A new level for the simulation of fluids and gases

Adam Moravanszky, a recognised figure as a senior director of simulation technology at Nvidia, emphatically indicated how GPU code integration within PhysX is a high-end application of the potential of CUDA and parallel programming models tailored for real-time physical simulation, a technology area that is during a time of intense and rapid development. For those who may not be familiar with this very specialised software, it is beneficial to explain that PhysX is a real-time physics simulation engine widely recognised for its ability to handle complex dynamics of colliding rigid bodies, deformable bodies reacting to external forces, and the fluid behaviour of gases and liquids.

Flow is especially interested in the simulation of fluids and gases and employs an effective approach based on the utilisation of sparse grids to deliver high visual and behavioural fidelity simulations. The union of these two tools places the hands of developers on the levers to create simulations of unprecedented detail and authenticity in a wide range of applications, ranging from the gaming and interactive entertainment industry to the field of sophisticated robotics and the virtual simulation of natural phenomena with unprecedented accuracy. Through such an initiating provocative declaration, Nvidia has not merely opened innovation across these crucial realms but potentially even exponentially.